Thread: question on forced-killing the postmaster

question on forced-killing the postmaster

From
shadowbox@punkass.com
Date:
why is it bad to kill -9 the postmaster? any links? i'm wondering why
is it bad...

percy

Re: question on forced-killing the postmaster

From
Tom Lane
Date:
shadowbox@punkass.com writes:
> why is it bad to kill -9 the postmaster?

Because then you end up with unreclaimed shared memory and semaphores.
Do this a few times and you'll run into the kernel limits on how much
of either you can have, and then be unable to start a fresh postmaster
without manual cleanup (or rebooting).

            regards, tom lane

Re: question on forced-killing the postmaster

From
Bruno Wolff III
Date:
On Tue, Apr 20, 2004 at 15:42:35 +0800,
  shadowbox@punkass.com wrote:
> why is it bad to kill -9 the postmaster? any links? i'm wondering why
> is it bad...

There might be more detail in the FAQ, but what I remember is that the
main issue is that shared memory doesn't necessarily get cleaned up
properly. You don't have to worry about losing committed transactions.

Re: question on forced-killing the postmaster

From
Bruce Momjian
Date:
shadowbox@punkass.com wrote:
> why is it bad to kill -9 the postmaster? any links? i'm wondering why
> is it bad...

The postmaster gets no way to shut down cleanly.  It just isn't as good
as using a normal kill and waiting a few seconds.  This is true for
killing most unix processes.  Use kill -9 only when other kills fail.

--
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 359-1001
  +  If your life is a hard drive,     |  13 Roberts Road
  +  Christ can be your backup.        |  Newtown Square, Pennsylvania 19073